compliant mechanism design

This work addresses the challenges of excessive actuation dimensionality and complex shape control in reconfigurable tendon-driven continuum robots, which arise from tendon rerouting. To circumvent the need for intricate dynamic modeling, the authors propose an actuation space reduction method that maps the backbone configuration into an intermediate curvature–torsion space to identify critical actuation disks. By integrating a rerouting mechanism based on actively rotated spacer disks and employing a proximal-to-distal staged shape-matching strategy, the approach effectively approximates the global configuration while enabling fine distal adjustments. This methodology significantly enhances the intuitiveness and efficiency of shape control without relying on complex kinematic or dynamic models.

This work proposes a tendon-driven continuum robot featuring a tapered flexible backbone fabricated from thermoplastic polyurethane (TPU), addressing limitations of conventional designs—such as high cost, poor customizability, and the difficulty of simultaneously achieving high curvature and distal compliance—in flexible manipulation tasks. An integrated electronic base enables precise tendon tension control and sensing. For the first time, the spatially varying tapered cross-section is explicitly incorporated into a forward dynamic-static model based on Cosserat rod theory to capture the influence of geometric tapering on stiffness distribution. Leveraging fused deposition modeling 3D printing and parametric CAD design, the system achieves low-cost, rapid assembly, and high customizability. Experimental calibration demonstrates centimeter-level shape prediction accuracy, and successful teleoperated endoscopic grasper tasks validate the robot’s efficacy in complex flexible manipulation scenarios.

To address the challenge of achieving both high load capacity and adaptive compliance in tendon-driven underactuated fingers, this paper proposes a compact, single-actuator design featuring full-joint mechanical coupling. A novel fixed-ratio synchronous tendon routing mechanism enables predictable stiffness and underactuated kinematic constraints while ensuring whole-finger unified actuation. We develop a static and kinematic model incorporating tendon elasticity and validate it experimentally using a 3D-printed prototype: under a 3 kg fingertip load, the finger achieves a stiffness of 1.2×10³ N/m, with deformation prediction error of only 1.0 mm (0.322% of finger length). This design significantly reduces complexity and weight in multi-fingered robotic hands; integrated into a five-fingered hand, it successfully accomplishes stable, adaptive grasping of diverse objects.

This study addresses the challenge that forward statics of multi-segment tendon-driven continuum robots in three-dimensional space typically rely on iterative solutions for nonlinear equilibrium. We propose a non-iterative, closed-form approach that adopts the centerline and cumulative twist as generalized coordinates. Based on variational principles, an explicit mapping from applied forces to Cartesian configurations is derived, replacing iterative computation with segment-wise propagation while preserving axial deformation, variable stiffness characteristics, and zero-equilibrium material twist constraints. Experimental results demonstrate that the proposed model achieves tip position errors below 1e-5 with a single computation time of approximately 3 microseconds, yielding over a 3000-fold speedup compared to baseline models. This work successfully unifies extremely high accuracy with real-time computational efficiency.

To address the conflicting challenges of low kinematic fidelity, insufficient rotational stiffness, and significant parasitic motion in large-angle (±15°) flexible crossed-hinge mechanisms, this paper proposes a static-dynamic-driven multi-objective optimization design methodology. We innovatively integrate rapid Euler–Bernoulli beam modeling with high-fidelity 3D ANSYS finite-element refinement to establish an interpretable hybrid modeling framework. Coupled with the NSGA-II algorithm, this approach efficiently explores the high-dimensional design parameter space and yields a Pareto-optimal solution set. The optimized configuration achieves motion error <0.5° over ±15° rotation, enhances rotational stiffness by 3.2×, and suppresses parasitic displacement by 87%, substantially outperforming conventional designs. This work provides both theoretical foundations and an engineering paradigm for high-performance compliant mechanisms.

This study addresses the inherent trade-off between joint torque and arm thickness in robotic manipulators by proposing a multi-objective optimization framework based on the NSGA-II algorithm. Through the co-optimization of tendon routing, pulley configurations, and attachment points, along with the introduction of a strategic shortcut mechanism to extend effective moment arms, the proposed method simultaneously maximizes torque output while minimizing structural thickness. The research reveals non-trivial design patterns that transcend conventional intuition and yields a Pareto-optimal solution set. These findings provide systematic engineering guidance for the design of compact, high-torque robotic arms, demonstrating that counterintuitive mechanical configurations can effectively reconcile competing performance objectives in manipulator design.

本文提出MuJoCable，通过优化路径算法和引入单向轴向定律等方法，在MuJoCo中添加了简化的缆绳传动模型，解决了肌腱驱动机器人中的摩擦和力传递问题。

This study addresses the low pose reproduction accuracy in underactuated tendon-driven hands caused by tendon elasticity. We propose the H-PAC modular robotic hand and a hierarchical control framework to overcome this limitation. By integrating a control-oriented sparse analytical model with a mechanical elasticity compensation mechanism and ESP32-based synchronous servo control, sensorless high-precision pose regulation is achieved. This approach effectively suppresses joint errors without requiring task-specific retuning. Experimental results demonstrate that the mean absolute error (MAE) of joint angle prediction remains below 0.23°, while the index finger DIP joint error is significantly reduced from 1.15° to 0.18°. These findings validate the proposed method’s effectiveness in achieving precision control for underactuated systems.

This study addresses the challenge of rapidly customizing stiffness in compliant mechanisms due to geometric constraints and stiffness coupling. A Lego-like stackable planar compliant module design method is proposed, establishing a unified model through a novel modular stiffness configuration framework to achieve stiffness decoupling and flexible reconfiguration. The optimization employs a genetic algorithm for module configuration search combined with sequential quadratic programming for parameter refinement. Experimental results demonstrate that simulated stiffness deviations remain below 6.5%. Furthermore, a developed compliant wrist prototype achieves approximately 15° angular compliance alongside prescribed stiffness characteristics during high-speed motion, validating the effectiveness of the proposed approach.

This study addresses the challenge of efficiently and accurately obtaining complete Cartesian backbone geometry for planning and control in tendon-driven continuum robots by proposing a simplified Cartesian computational framework. The method achieves efficient equilibrium configuration solving along predefined trajectories through Taylor-Galerkin model order reduction and stable residual propagation. By introducing offline moment vectors and an analytical residual correction mechanism, it eliminates the need for online spatial integration while significantly suppressing propagation drift. Furthermore, global position field representation is realized via analytical differentiation combined with a fixed-dimensional linear solver. Experimental results demonstrate that the average update time is only 0.508 milliseconds—an elevenfold speedup over conventional methods—with tip positioning errors as low as 0.28%, thereby achieving high-precision, real-time full-shape prediction.